Passage: "The recent surge in 'algorithmic bias' cases has brought renewed attention to the ethical implications of artificial intelligence (AI) in legal decision-making. While proponents argue that AI can increase efficiency and reduce human error, critics point to instances where AI systems have perpetuated or even amplified existing societal biases. These systems, often trained on historical data, can inadvertently learn and embed discriminatory patterns, leading to unfair outcomes in areas such as sentencing, bail recommendations, and even predictive policing. The challenge lies not in the algorithms themselves, which are mathematical constructs, but in the data they consume and the human assumptions embedded in their design. Addressing algorithmic bias requires a multi-faceted approach, including diverse data sets, transparent algorithmic design, and rigorous independent auditing." Which of the following would the author most likely consider a necessary step in mitigating algorithmic bias?

1. ALimiting the use of AI in all legal decision-making processes.
2. BImplementing independent audits of AI systems used in legal contexts.
3. CReplacing all human decision-makers with AI systems to eliminate bias.
4. DFocusing solely on improving the mathematical constructs of algorithms.